AI Engineer (m/f/n)
⚲ Warszawa
26 880 - 29 400 PLN netto (B2B)
Wymagania
- Python
Opis stanowiska
urrently, for one of our Partners, we are looking for an experienced AI Engineer (m/f/n) to support a new project delivered for a US-based enterprise environment.
The project focuses on building agentic workflows for a production LLM and deterministic Python-based tax pipeline.
Scope of duties:
• Building planning and rollover agents for a production LLM and deterministic Python-based tax pipeline.
• Integrating AI agents with enterprise applications via REST APIs.
• Building and consuming agent tools using Model Context Protocol / MCP.
• Delivering an agent-to-agent / A2A-pluggable agent for an enhanced graph-based orchestrator.
• Working with LangGraph-based orchestration, including workflow nodes, human-in-the-loop interrupts, and durable checkpointing.
• Implementing structured outputs, tool calling, and multi-agent/sub-agent patterns.
• Optimizing agent workflows for cost and latency.
• Developing automated tests for async Python services and agent workflows.
• Supporting integration with enterprise authentication mechanisms such as OAuth2 / SSO.
• Cooperating with technical stakeholders in an international, English-speaking environment.
Requirements
• Software engineering experience: commercial experience in backend/software development.
• Agentic workflows development: at least 1 year.
• Enterprise API integration experience: practical experience with real production systems.
• Python development with asyncio.
• Typed data validation and settings management with Pydantic.
• Async web/API frameworks, especially FastAPI.
• Automated testing with pytest and pytest-asyncio.
• Building LLM agents, including tool calling, structured outputs, multi-agent/sub-agent patterns, and cost/latency control.
• Experience with a mainstream LLM agent framework, e.g. Claude Agent SDK.
• Building and consuming agent tool servers using Model Context Protocol / MCP.
• Graph-based stateful orchestration with LangGraph, including workflow nodes, human-in-the-loop interrupts, and durable checkpointing.
• Experience with agent-to-agent / A2A protocols.
• Enterprise REST integration with OAuth2 / SSO, e.g. Azure AD.
• Distributed in-memory data stores, e.g. Redis / Redis Stack.
• Containerization with Docker.
• English: C1.
Nice to have
• Observability experience with OpenTelemetry, Datadog, or Langfuse LLM tracing.
• Cloud services experience, especially Azure Blob Storage or managed search.
• Experience with LLM evaluations, including LLM-as-judge.
• Spec-driven development experience.
• Tax or accounting domain knowledge.
• Previous experience working within large enterprise ecosystems.
Offer
• Contract: B2B via SHIMI Poland.
• Type of work: remote.
• Rate: up to 175 PLN/h net + VAT.
• Private medical care and sportscard.
The project focuses on building agentic workflows for a production LLM and deterministic Python-based tax pipeline.
Scope of duties:
• Building planning and rollover agents for a production LLM and deterministic Python-based tax pipeline.
• Integrating AI agents with enterprise applications via REST APIs.
• Building and consuming agent tools using Model Context Protocol / MCP.
• Delivering an agent-to-agent / A2A-pluggable agent for an enhanced graph-based orchestrator.
• Working with LangGraph-based orchestration, including workflow nodes, human-in-the-loop interrupts, and durable checkpointing.
• Implementing structured outputs, tool calling, and multi-agent/sub-agent patterns.
• Optimizing agent workflows for cost and latency.
• Developing automated tests for async Python services and agent workflows.
• Supporting integration with enterprise authentication mechanisms such as OAuth2 / SSO.
• Cooperating with technical stakeholders in an international, English-speaking environment.
Requirements
• Software engineering experience: commercial experience in backend/software development.
• Agentic workflows development: at least 1 year.
• Enterprise API integration experience: practical experience with real production systems.
• Python development with asyncio.
• Typed data validation and settings management with Pydantic.
• Async web/API frameworks, especially FastAPI.
• Automated testing with pytest and pytest-asyncio.
• Building LLM agents, including tool calling, structured outputs, multi-agent/sub-agent patterns, and cost/latency control.
• Experience with a mainstream LLM agent framework, e.g. Claude Agent SDK.
• Building and consuming agent tool servers using Model Context Protocol / MCP.
• Graph-based stateful orchestration with LangGraph, including workflow nodes, human-in-the-loop interrupts, and durable checkpointing.
• Experience with agent-to-agent / A2A protocols.
• Enterprise REST integration with OAuth2 / SSO, e.g. Azure AD.
• Distributed in-memory data stores, e.g. Redis / Redis Stack.
• Containerization with Docker.
• English: C1.
Nice to have
• Observability experience with OpenTelemetry, Datadog, or Langfuse LLM tracing.
• Cloud services experience, especially Azure Blob Storage or managed search.
• Experience with LLM evaluations, including LLM-as-judge.
• Spec-driven development experience.
• Tax or accounting domain knowledge.
• Previous experience working within large enterprise ecosystems.
Offer
• Contract: B2B via SHIMI Poland.
• Type of work: remote.
• Rate: up to 175 PLN/h net + VAT.
• Private medical care and sportscard.
🔍 Dekoder Ogłoszenia
🟡
experienced AI Engineer
Poszukujemy kogoś, kto ma już praktyczne doświadczenie w tworzeniu i wdrażaniu systemów opartych na AI, a nie tylko teoretyczną wiedzę.
🟡
new project delivered for a US-based enterprise environment
Projekt jest nowy, co może oznaczać brak utartych ścieżek i potrzebę budowania rozwiązań od podstaw, ale też potencjalnie większą swobodę.
🟡
production LLM
System będzie działał w środowisku produkcyjnym, co wymaga wysokiej stabilności, wydajności i niezawodności.
🟡
agentic workflows
Praca będzie polegać na tworzeniu i zarządzaniu złożonymi, autonomicznie działającymi procesami opartymi na AI.
🟡
at least 1 year of Agentic workflows development
Wymagane jest konkretne, udokumentowane doświadczenie w tej specyficznej dziedzinie, a nie tylko ogólne pojęcie.